web2md 0.1.2

A tool that fetches web pages and returns them as Markdown for MCP token efficiency
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
use anyhow::Result;
use url::Url;

/// Case-insensitive byte search. Returns byte position of `needle` in `haystack`.
pub(crate) fn find_ci(haystack: &str, needle: &str) -> Option<usize> {
    if needle.is_empty() {
        return Some(0);
    }
    let needle_lower: Vec<u8> = needle.bytes().map(|b| b.to_ascii_lowercase()).collect();
    let h = haystack.as_bytes();
    if needle_lower.len() > h.len() {
        return None;
    }
    'outer: for i in 0..=h.len() - needle_lower.len() {
        for (j, &n) in needle_lower.iter().enumerate() {
            if h[i + j].to_ascii_lowercase() != n {
                continue 'outer;
            }
        }
        return Some(i);
    }
    None
}

/// Normalize a Markdown block for deduplication comparison.
/// Collapses whitespace and trims, so blocks differing only in spacing are treated as duplicates.
fn normalize_block(block: &str) -> String {
    block.split_whitespace().collect::<Vec<_>>().join(" ")
}

/// Extract language from a `<code class="language-xxx">` tag's class attribute.
/// Returns None if no language class is found.
fn extract_language_class(tag: &str) -> Option<String> {
    let class_pos = find_ci(tag, "class=")?;
    let after = &tag[class_pos + 6..];
    let mut i = 0;
    while i < after.len() && after.as_bytes()[i].is_ascii_whitespace() {
        i += 1;
    }
    let quote = *after.as_bytes().get(i)? as char;
    if quote != '"' && quote != '\'' {
        return None;
    }
    let val_start = i + 1;
    let val_end = after[val_start..].find(quote)? + val_start;
    let class_val = &after[val_start..val_end];
    for cls in class_val.split_whitespace() {
        if let Some(lang) = cls.strip_prefix("language-") {
            if !lang.is_empty() {
                return Some(lang.to_string());
            }
        }
    }
    None
}

/// Convert raw HTML to clean Markdown.
/// Strips scripts, styles, and non-essential markup to minimize token output.
pub struct PageToMarkdown;

impl PageToMarkdown {
    /// Convert HTML string to Markdown.
    /// When `include_images` is false, strips `<img>` tags to reduce token output.
    /// When `main_content` is true, extracts only the content of `<article>`, `<main>`, or `[role="main"]` elements.
    /// When `exclude_selectors` is non-empty, removes HTML elements matching the given CSS-like selectors (`.class` or `#id`) before conversion.
    pub fn convert(html: &str, include_images: bool, keep_header: bool, main_content: bool, exclude_selectors: &[String]) -> Result<String> {
        let original_html = html.to_string();
        let html = if main_content {
            Self::extract_main_content(&original_html)
        } else {
            original_html.clone()
        };
        let html = Self::strip_scripts_and_styles(&html);
        let html = Self::strip_iframe_tags(&html);
        let html = Self::strip_noise_tags(&html, keep_header);
        let html = Self::strip_html_comments(&html);
        let html = Self::strip_by_selectors(&html, exclude_selectors);
        let languages = Self::extract_code_languages(&html);
        let html = if include_images { html } else { Self::strip_img_tags(&html) };
        let md = html2md::parse_html(&html);
        let md = Self::inject_code_languages(&md, &languages);
        let md = Self::deduplicate_blocks(&md);
        let md = Self::clean(&md);

        // Append extracted comments for forum/thread pages
        if let Some(comments) = Self::extract_comments(&original_html) {
            Ok(format!("{}\n\n{}", md, comments))
        } else {
            Ok(md)
        }
    }

    /// Convert relative URLs in Markdown links to absolute URLs using the given base URL.
    /// Processes `[text](url)` and `![alt](url)` patterns, leaving already-absolute URLs unchanged.
    pub fn absolutize_links(md: &str, base_url: &str) -> String {
        let base = match Url::parse(base_url) {
            Ok(u) => u,
            Err(_) => return md.to_string(),
        };

        let mut result = String::with_capacity(md.len());
        let mut i = 0;

        while i < md.len() {
            // Look for `](` pattern which indicates a Markdown link
            let bytes = md.as_bytes();
            if bytes[i] == b']' && i + 1 < md.len() && bytes[i + 1] == b'(' {
                // Find the closing ')'
                if let Some(close) = md[i + 2..].find(')') {
                    let url_start = i + 2;
                    let url_end = i + 2 + close;
                    let raw_url = &md[url_start..url_end];

                    // Skip empty URLs and anchor-only links
                    if raw_url.is_empty() || raw_url.starts_with('#') {
                        result.push_str(&md[i..=url_end]);
                        i = url_end + 1;
                        continue;
                    }

                    // Try to resolve the URL against the base
                    if let Ok(absolved) = base.join(raw_url) {
                        result.push_str("](");
                        result.push_str(absolved.as_str());
                        result.push(')');
                        i = url_end + 1;
                        continue;
                    }

                    // If resolution fails, keep the original
                    result.push_str(&md[i..=url_end]);
                    i = url_end + 1;
                    continue;
                }
            }
            // Push one character (UTF-8 safe) from the current position
            let ch = md[i..].chars().next().unwrap();
            result.push(ch);
            i += ch.len_utf8();
        }

        result
    }

    /// Remove `<img>` tags (case-insensitive). Self-closing (`/>`) or plain (`>`) both handled.
    fn strip_img_tags(html: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        while i < html.len() {
            if let Some(start) = find_ci(&html[i..], "<img") {
                let start = i + start;
                if let Some(end) = html[start..].find('>') {
                    out.push_str(&html[i..start]);
                    i = start + end + 1;
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Remove `<script>` and `<style>` blocks (case-insensitive, non-greedy)
    fn strip_scripts_and_styles(html: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        while i < html.len() {
            let script_start = find_ci(&html[i..], "<script").map(|p| i + p);
            let style_start = find_ci(&html[i..], "<style").map(|p| i + p);
            let next = match (script_start, style_start) {
                (Some(s), Some(st)) => Some((s.min(st), s == s.min(st))),
                (Some(s), None) => Some((s, true)),
                (None, Some(st)) => Some((st, false)),
                (None, None) => None,
            };
            if let Some((start, is_script)) = next {
                let close = if is_script { "</script>" } else { "</style>" };
                if let Some(end) = find_ci(&html[start..], close) {
                    out.push_str(&html[i..start]);
                    i = start + end + close.len();
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Remove non-content HTML tags: `<nav>`, `<footer>`, `<aside>`, `<noscript>`, `<form>`.
    /// These are navigation, structural, or interactive elements that add noise to Markdown output.
    fn strip_noise_tags(html: &str, keep_header: bool) -> String {
        let mut tags = vec![
            ("nav", "</nav>"),
            ("footer", "</footer>"),
            ("aside", "</aside>"),
            ("noscript", "</noscript>"),
            ("form", "</form>"),
        ];
        if !keep_header {
            tags.push(("header", "</header>"));
        }
        let mut result = html.to_string();
        for (open, close) in &tags {
            result = Self::strip_tag_pair(&result, open, close);
        }
        result
    }

    /// Generic case-insensitive removal of `<tag>...</tag>` blocks.
    fn strip_tag_pair(html: &str, tag: &str, close_tag: &str) -> String {
        let open = format!("<{}", tag);
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        while i < html.len() {
            if let Some(start) = find_ci(&html[i..], &open) {
                let start = i + start;
                if let Some(end) = find_ci(&html[start..], close_tag) {
                    out.push_str(&html[i..start]);
                    i = start + end + close_tag.len();
                    continue;
                }
                // unclosed: skip to next `>` to avoid eating the rest of the document
                if let Some(gt) = html[start..].find('>') {
                    out.push_str(&html[i..start]);
                    i = start + gt + 1;
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Remove HTML comments `<!-- ... -->` (case-insensitive on delimiters).
    fn strip_html_comments(html: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        while i < html.len() {
            if let Some(start) = find_ci(&html[i..], "<!--") {
                let start = i + start;
                if let Some(end) = find_ci(&html[start..], "-->") {
                    out.push_str(&html[i..start]);
                    i = start + end + 3;
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Remove HTML elements matching CSS-like selectors from the HTML.
    /// Supports `.classname` (class selector) and `#id` (ID selector).
    /// Multiple selectors can be provided; each is applied in order.
    fn strip_by_selectors(html: &str, selectors: &[String]) -> String {
        let mut result = html.to_string();
        for selector in selectors {
            let trimmed = selector.trim();
            if let Some(class_name) = trimmed.strip_prefix('.') {
                if !class_name.is_empty() {
                    result = Self::strip_elements_by_attr(&result, "class", class_name);
                }
            } else if let Some(id_val) = trimmed.strip_prefix('#') {
                if !id_val.is_empty() {
                    result = Self::strip_elements_by_attr(&result, "id", id_val);
                }
            }
        }
        result
    }

    /// Remove HTML elements where an attribute contains the given value.
    /// Matches `class="foo bar"` for value "foo" (word-boundary match),
    /// and `id="exact"` for value "exact" (exact match).
    fn strip_elements_by_attr(html: &str, attr: &str, value: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        let lower_attr = format!("{}=\"", attr);

        while i < html.len() {
            // Find the next opening tag
            if let Some(rel_start) = html[i..].find('<') {
                let tag_start = i + rel_start;
                // Find the end of this opening tag
                if let Some(rel_end) = html[tag_start..].find('>') {
                    let tag_end = tag_start + rel_end;
                    let tag_content = &html[tag_start..=tag_end];
                    let tag_lower = tag_content.to_ascii_lowercase();

                    // Check if this tag has the attribute we're looking for
                    if let Some(attr_pos) = find_ci(&tag_lower, &lower_attr) {
                        let val_start = attr_pos + lower_attr.len();
                        if let Some(quote_end) = tag_lower[val_start..].find('"') {
                            let attr_val = &tag_lower[val_start..val_start + quote_end];
                            let matches = if attr == "id" {
                                attr_val == value.to_ascii_lowercase()
                            } else {
                                attr_val.split_whitespace().any(|w| w == value.to_ascii_lowercase().as_str())
                            };

                            if matches {
                                // Determine the tag name
                                let after_lt = &tag_content[1..];
                                let tag_name: String = after_lt
                                    .chars()
                                    .take_while(|c| c.is_alphanumeric())
                                    .collect();

                                // Copy everything before this tag
                                out.push_str(&html[i..tag_start]);

                                if tag_name.is_empty() {
                                    i = tag_end + 1;
                                    continue;
                                }

                                let void_tags = ["br", "hr", "img", "input", "meta", "link", "area", "base", "col", "embed", "source", "track", "wbr"];
                                if void_tags.contains(&tag_name.as_str()) {
                                    i = tag_end + 1;
                                    continue;
                                }

                                // Find the matching close tag
                                let close_tag = format!("</{}", tag_name);
                                if let Some(close_pos) = find_ci(&html[tag_end + 1..], &close_tag) {
                                    let close_start = tag_end + 1 + close_pos;
                                    if let Some(close_end) = html[close_start..].find('>') {
                                        i = close_start + close_end + 1;
                                        continue;
                                    }
                                }

                                // No close tag found, just remove the opening tag
                                i = tag_end + 1;
                                continue;
                            }
                        }
                    }

                    // Tag doesn't match — copy it and advance past it
                    out.push_str(&html[i..tag_end + 1]);
                    i = tag_end + 1;
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Extract language annotations from `<code class="language-xxx">` tags, in document order.
    /// Returns a list of languages corresponding to code blocks in the HTML.
    fn extract_code_languages(html: &str) -> Vec<String> {
        let mut languages = Vec::new();
        let mut i = 0;
        while i < html.len() {
            if let Some(pos) = find_ci(&html[i..], "<code") {
                let pos = i + pos;
                if let Some(gt) = html[pos..].find('>') {
                    let tag = &html[pos..=pos + gt];
                    if let Some(lang) = extract_language_class(tag) {
                        languages.push(lang);
                    }
                    i = pos + gt + 1;
                    continue;
                }
            }
            break;
        }
        languages
    }

    /// Inject language annotations into fenced code blocks that lack them.
    /// Pairs with `extract_code_languages` — languages are matched in order.
    fn inject_code_languages(md: &str, languages: &[String]) -> String {
        if languages.is_empty() {
            return md.to_string();
        }
        let mut result = String::with_capacity(md.len());
        let mut lang_idx = 0;
        let lines: Vec<&str> = md.lines().collect();
        let mut i = 0;
        while i < lines.len() {
            let line = lines[i];
            let trimmed = line.trim();
            if trimmed == "```" && lang_idx < languages.len() {
                result.push_str(&format!("```{}", languages[lang_idx]));
                lang_idx += 1;
            } else {
                result.push_str(line);
            }
            if i + 1 < lines.len() {
                result.push('\n');
            }
            i += 1;
        }
        result
    }

    /// Remove `<iframe>` tags including their closing tag (case-insensitive).
    fn strip_iframe_tags(html: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut i = 0;
        while i < html.len() {
            if let Some(start) = find_ci(&html[i..], "<iframe") {
                let start = i + start;
                if let Some(end) = find_ci(&html[start..], "</iframe>") {
                    out.push_str(&html[i..start]);
                    i = start + end + "</iframe>".len();
                    continue;
                }
                // self-closing or unclosed: find next >
                if let Some(end) = html[start..].find('>') {
                    out.push_str(&html[i..start]);
                    i = start + end + 1;
                    continue;
                }
            }
            out.push_str(&html[i..]);
            break;
        }
        out
    }

    /// Extract the main content area from HTML.
    /// Looks for `<article>`, `<main>`, or `<div role="main">` tags (in priority order).
    /// Returns the inner HTML of the first match. If none found, returns the full HTML.
    fn extract_main_content(html: &str) -> String {
        for (tag, close_tag) in [("article", "</article>"), ("main", "</main>")] {
            if let Some(start) = find_ci(html, &format!("<{}", tag)) {
                if let Some(gt) = html[start..].find('>') {
                    let content_start = start + gt + 1;
                    if let Some(end) = find_ci(&html[content_start..], close_tag) {
                        return html[content_start..content_start + end].to_string();
                    }
                }
            }
        }
        // Look for <div role="main">
        let mut i = 0;
        while i < html.len() {
            if let Some(pos) = find_ci(&html[i..], "<div") {
                let pos = i + pos;
                if let Some(gt) = html[pos..].find('>') {
                    let tag = &html[pos..=pos + gt];
                    if find_ci(tag, "role=\"main\"").is_some() || find_ci(tag, "role='main'").is_some() {
                        let content_start = pos + gt + 1;
                        if let Some(end) = find_ci(&html[content_start..], "</div>") {
                            return html[content_start..content_start + end].to_string();
                        }
                    }
                    i = pos + gt + 1;
                    continue;
                }
            }
            break;
        }
        // Readability fallback: score top-level <div> elements by text density
        Self::readability_extract(html)
    }

    /// Readability-based content extraction.
    /// Scores top-level `<div>` and `<section>` blocks by text density and link density.
    /// Falls back to paragraph-level scoring if no block-level candidate scores well.
    /// Returns the inner HTML of the highest-scoring block, or the full HTML if no suitable block is found.
    fn readability_extract(html: &str) -> String {
        let candidates = Self::find_top_level_blocks(html);
        if candidates.is_empty() {
            return Self::paragraph_readability_extract(html);
        }

        let mut best_score = 0i64;
        let mut best_html = None;

        for (content, _tag) in &candidates {
            let text_len = Self::count_text_chars(content) as i64;
            let link_text_len = Self::count_link_text_chars(content) as i64;
            // Score: text content minus link text (navigation penalized)
            let score = text_len - link_text_len;
            if score > best_score {
                best_score = score;
                best_html = Some(content.clone());
            }
        }

        // Only use readability result if the best block has meaningful content (>100 chars of non-link text)
        if best_score > 100 {
            best_html.unwrap_or_else(|| html.to_string())
        } else {
            Self::paragraph_readability_extract(html)
        }
    }

    /// Paragraph-level readability extraction.
    /// Finds all `<p>` blocks, scores them by text length, and extracts the contiguous window
    /// of paragraphs with the highest combined text density.
    /// Falls back to the full HTML if no good paragraph cluster is found.
    fn paragraph_readability_extract(html: &str) -> String {
        let paragraphs = Self::find_paragraph_blocks(html);
        if paragraphs.is_empty() {
            return html.to_string();
        }

        // Score each paragraph by text length (longer = more likely content)
        let scores: Vec<usize> = paragraphs
            .iter()
            .map(|(content, _, _)| Self::count_text_chars(content))
            .collect();

        // Find the contiguous window with the highest total text score.
        // Use a sliding window of up to 5 paragraphs to find the densest cluster.
        let window_size = 5.min(paragraphs.len());
        let mut best_start = 0;
        let mut best_score = 0usize;

        for start in 0..=paragraphs.len() - window_size {
            let score: usize = scores[start..start + window_size].iter().sum();
            if score > best_score {
                best_score = score;
                best_start = start;
            }
        }

        // Only use paragraph extraction if the best window has meaningful content
        if best_score < 100 {
            return html.to_string();
        }

        let end = (best_start + window_size).min(paragraphs.len());
        let html_start = paragraphs[best_start].1;
        let html_end = paragraphs[end - 1].2;
        html[html_start..html_end].to_string()
    }

    /// Find all `<p>` blocks in the HTML.
    /// Returns a list of (inner_html, start_byte, end_byte) tuples.
    fn find_paragraph_blocks(html: &str) -> Vec<(String, usize, usize)> {
        let mut blocks = Vec::new();
        let mut i = 0;
        while i < html.len() {
            if let Some(pos) = find_ci(&html[i..], "<p") {
                let pos = i + pos;
                // Ensure this is a <p> tag, not <pre>, <param>, etc.
                let after = &html[pos + 2..];
                let next_char = after.chars().next();
                if next_char != Some(' ') && next_char != Some('>') && next_char != Some('\t') && next_char != Some('\n') && next_char != Some('\r') {
                    i = pos + 2;
                    continue;
                }
                if let Some(gt) = html[pos..].find('>') {
                    let content_start = pos + gt + 1;
                    if let Some(end) = find_ci(&html[content_start..], "</p>") {
                        let end_abs = content_start + end;
                        blocks.push((
                            html[content_start..end_abs].to_string(),
                            pos,
                            end_abs + "</p>".len(),
                        ));
                        i = end_abs + "</p>".len();
                        continue;
                    }
                    // Unclosed <p>: take everything until the next <p> or block-level tag
                    i = content_start;
                } else {
                    break;
                }
            } else {
                break;
            }
        }
        blocks
    }

    /// Find top-level `<div>` and `<section>` blocks in HTML.
    /// Returns a list of (inner_html, tag_name) pairs, respecting nesting depth.
    fn find_top_level_blocks(html: &str) -> Vec<(String, &'static str)> {
        let mut blocks = Vec::new();
        for tag in &["div", "section"] {
            let open = format!("<{}", tag);
            let close = format!("</{}>", tag);
            let mut i = 0;
            while i < html.len() {
                if let Some(pos) = find_ci(&html[i..], &open) {
                    let pos = i + pos;
                    if let Some(gt) = html[pos..].find('>') {
                        let content_start = pos + gt + 1;
                        // Find matching close tag, respecting nesting
                        if let Some(end) = Self::find_matching_close(html, content_start, &open, &close) {
                            blocks.push((html[content_start..end].to_string(), *tag));
                            i = end + close.len();
                            continue;
                        }
                    }
                    i = pos + 1;
                } else {
                    break;
                }
            }
        }
        blocks
    }

    /// Find the matching close tag for an HTML block, respecting nesting depth.
    fn find_matching_close(html: &str, start: usize, open: &str, close: &str) -> Option<usize> {
        let mut depth = 1;
        let mut i = start;
        while i < html.len() {
            let rest = &html[i..];
            let next_open = find_ci(rest, open).map(|p| i + p);
            let next_close = find_ci(rest, close).map(|p| i + p);
            match (next_open, next_close) {
                (Some(o), Some(c)) if o < c => {
                    depth += 1;
                    i = o + open.len();
                }
                (_, Some(c)) => {
                    depth -= 1;
                    if depth == 0 {
                        return Some(c);
                    }
                    i = c + close.len();
                }
                _ => return None,
            }
        }
        None
    }

    /// Count visible text characters in HTML (excluding tag content and whitespace).
    fn count_text_chars(html: &str) -> usize {
        let mut count = 0;
        let mut in_tag = false;
        for c in html.chars() {
            if c == '<' {
                in_tag = true;
            } else if c == '>' {
                in_tag = false;
            } else if !in_tag && !c.is_ascii_whitespace() {
                count += 1;
            }
        }
        count
    }

    /// Count text characters inside `<a>` tags (used as a proxy for link/navigation density).
    fn count_link_text_chars(html: &str) -> usize {
        let mut count = 0;
        let mut i = 0;
        while i < html.len() {
            if let Some(pos) = find_ci(&html[i..], "<a") {
                let pos = i + pos;
                // Ensure this is an <a> tag, not <article>, <aside>, etc.
                let after = &html[pos + 2..];
                let next_char = after.chars().next();
                if next_char != Some(' ') && next_char != Some('>') && next_char != Some('\t') && next_char != Some('\n') && next_char != Some('\r') {
                    i = pos + 2;
                    continue;
                }
                if let Some(gt) = html[pos..].find('>') {
                    let content_start = pos + gt + 1;
                    if let Some(end) = find_ci(&html[content_start..], "</a>") {
                        let end_abs = content_start + end;
                        let link_text = &html[content_start..end_abs];
                        count += Self::count_text_chars(link_text);
                        i = end_abs + "</a>".len();
                        continue;
                    }
                }
                i = pos + 1;
            } else {
                break;
            }
        }
        count
    }

    /// Remove duplicate paragraph-level blocks from Markdown.
    /// Blocks are separated by blank lines. Only substantial blocks (>20 chars of normalized text)
    /// are deduplicated — short blocks like headings or single words are kept.
    fn deduplicate_blocks(md: &str) -> String {
        use std::collections::HashSet;

        let mut seen: HashSet<String> = HashSet::new();
        let mut result = String::with_capacity(md.len());
        let mut current_block = String::new();

        for line in md.lines() {
            if line.trim().is_empty() {
                if !current_block.is_empty() {
                    let normalized = normalize_block(&current_block);
                    if normalized.len() <= 20 || seen.insert(normalized) {
                        result.push_str(&current_block);
                        result.push('\n');
                    }
                    current_block.clear();
                }
                result.push('\n');
            } else {
                current_block.push_str(line);
                current_block.push('\n');
            }
        }

        if !current_block.is_empty() {
            let normalized = normalize_block(&current_block);
            if normalized.len() <= 20 || seen.insert(normalized) {
                result.push_str(&current_block);
            }
        }

        result
    }

    /// Post-process: collapse excessive whitespace and trim
    fn clean(md: &str) -> String {
        let mut out = String::new();
        let mut blank_lines = 0;

        for line in md.lines() {
            let trimmed = line.trim();
            if trimmed.is_empty() {
                blank_lines += 1;
                if blank_lines <= 2 {
                    out.push('\n');
                }
            } else {
                blank_lines = 0;
                out.push_str(trimmed);
                out.push('\n');
            }
        }
        out.trim().to_string()
    }

    /// Detect if a page looks like a forum/thread with comments.
    /// Returns true if multiple comment-like containers are found.
    fn looks_like_forum_page(html: &str) -> bool {
        let comment_indicators = [
            "class=\"comment",
            "class='comment",
            "class=\"post-body",
            "class='post-body",
            "class=\"comment-body",
            "class='comment-body",
            "class=\"comment-content",
            "class='comment-content",
            "class=\"message-body",
            "class='message-body",
            "id=\"comment-",
            "id='comment-",
            "data-comment-id",
            "data-testid=\"comment",
        ];
        let mut count = 0;
        for indicator in &comment_indicators {
            count += find_ci(html, indicator).map(|_| 1).unwrap_or(0);
            if count >= 2 {
                return true;
            }
        }
        false
    }

    /// Extract comments from forum/thread pages.
    /// Detects comment containers, extracts author and text, and formats as Markdown.
    /// Returns None if no comments are found or the page doesn't look like a forum.
    fn extract_comments(html: &str) -> Option<String> {
        if !Self::looks_like_forum_page(html) {
            return None;
        }

        let comments = Self::find_comment_blocks(html);
        if comments.len() < 2 {
            return None;
        }

        let mut out = String::new();
        out.push_str("## Comments\n\n");
        for (author, text, depth) in &comments {
            let indent = "  ".repeat(*depth);
            if let Some(a) = author {
                out.push_str(&format!("{}**{}**:\n\n", indent, a));
            }
            let comment_md = html2md::parse_html(text);
            let comment_md = Self::clean(&comment_md);
            for line in comment_md.lines() {
                out.push_str(&format!("{}> {}\n", indent, line));
            }
            out.push('\n');
        }
        Some(out.trim().to_string())
    }

    /// Find comment blocks in HTML.
    /// Returns a list of (author, inner_html, nesting_depth) tuples.
    fn find_comment_blocks(html: &str) -> Vec<(Option<String>, String, usize)> {
        let mut blocks = Vec::new();
        let comment_patterns = [
            "class=\"comment",
            "class='comment",
            "id=\"comment-",
            "id='comment-",
            "data-testid=\"comment",
        ];

        let mut i = 0;
        while i < html.len() {
            let mut found = false;
            for pattern in &comment_patterns {
                if let Some(pos) = find_ci(&html[i..], pattern) {
                    let pos = i + pos;
                    // Find the enclosing tag start (walk back to '<')
                    let tag_start = html[..pos].rfind('<').unwrap_or(pos);
                    if let Some(gt) = html[tag_start..].find('>') {
                        let tag = &html[tag_start..=tag_start + gt];
                        let tag_name = Self::extract_tag_name(tag);
                        let close_tag = format!("</{}>", tag_name);
                        let content_start = tag_start + gt + 1;
                        if let Some(end) = Self::find_matching_close(
                            html,
                            content_start,
                            &format!("<{}", tag_name),
                            &close_tag,
                        ) {
                            let inner = &html[content_start..end];
                            // Check the container tag for data-author first, then inner HTML
                            let author = Self::extract_comment_author(tag)
                                .or_else(|| Self::extract_comment_author(inner));
                            let depth = Self::estimate_nesting_depth(html, tag_start);
                            let text = Self::extract_comment_text(inner);
                            if !text.is_empty() {
                                blocks.push((author, text, depth));
                            }
                            i = end + close_tag.len();
                            found = true;
                            break;
                        }
                    }
                }
            }
            if !found {
                break;
            }
        }
        blocks
    }

    /// Extract the tag name from an opening tag string like `<div class="...">`.
    fn extract_tag_name(tag: &str) -> String {
        let after_lt = &tag[1..];
        let end = after_lt
            .find(|c: char| c.is_ascii_whitespace() || c == '>' || c == '/')
            .unwrap_or(after_lt.len());
        after_lt[..end].to_string()
    }

    /// Extract author name from comment inner HTML.
    /// Looks for common author patterns: `<a class="author">`, `<span class="author">`,
    /// `data-author="..."`, `<cite>`, or `<address>`.
    fn extract_comment_author(html: &str) -> Option<String> {
        // data-author attribute
        if let Some(pos) = find_ci(html, "data-author=") {
            let after = &html[pos + 12..];
            if let Some(val) = Self::extract_quoted_value(after) {
                return Some(val);
            }
        }
        // <a class="author"> or <span class="author">
        for tag in &["<a", "<span", "<div", "<strong", "<b"] {
            let mut i = 0;
            while i < html.len() {
                if let Some(pos) = find_ci(&html[i..], tag) {
                    let pos = i + pos;
                    if let Some(gt) = html[pos..].find('>') {
                        let tag_str = &html[pos..=pos + gt];
                        if find_ci(tag_str, "author").is_some()
                            || find_ci(tag_str, "username").is_some()
                            || find_ci(tag_str, "user-name").is_some()
                        {
                            let content_start = pos + gt + 1;
                            let close = format!("</{}", &tag[1..]);
                            if let Some(end) = find_ci(&html[content_start..], &close) {
                                let author_html = &html[content_start..content_start + end];
                                let author = Self::strip_tags(author_html);
                                let author = author.trim().to_string();
                                if !author.is_empty() {
                                    return Some(author);
                                }
                            }
                        }
                        i = pos + gt + 1;
                        continue;
                    }
                }
                break;
            }
        }
        // <cite> or <address>
        for tag in &["<cite", "<address"] {
            if let Some(pos) = find_ci(html, tag) {
                if let Some(gt) = html[pos..].find('>') {
                    let content_start = pos + gt + 1;
                    let close = format!("</{}", &tag[1..]);
                    if let Some(end) = find_ci(&html[content_start..], &close) {
                        let author = Self::strip_tags(&html[content_start..content_start + end]);
                        let author = author.trim().to_string();
                        if !author.is_empty() {
                            return Some(author);
                        }
                    }
                }
            }
        }
        None
    }

    /// Extract the main text content from a comment block, excluding metadata.
    /// Looks for content containers first, then falls back to the full inner HTML.
    fn extract_comment_text(html: &str) -> String {
        let content_patterns = [
            "class=\"comment-body",
            "class='comment-body",
            "class=\"comment-content",
            "class='comment-content",
            "class=\"post-body",
            "class='post-body",
            "class=\"message-body",
            "class='message-body",
            "class=\"usertext-body",
            "class='usertext-body",
            "class=\"md",
            "class='md",
        ];
        for pattern in &content_patterns {
            if let Some(pos) = find_ci(html, pattern) {
                let tag_start = html[..pos].rfind('<').unwrap_or(pos);
                if let Some(gt) = html[tag_start..].find('>') {
                    let tag = &html[tag_start..=tag_start + gt];
                    let tag_name = Self::extract_tag_name(tag);
                    let close_tag = format!("</{}>", tag_name);
                    let content_start = tag_start + gt + 1;
                    if let Some(end) =
                        Self::find_matching_close(html, content_start, &format!("<{}", tag_name), &close_tag)
                    {
                        return html[content_start..end].to_string();
                    }
                }
            }
        }
        // Fallback: return the full inner HTML
        html.to_string()
    }

    /// Estimate nesting depth by counting ancestor comment containers.
    fn estimate_nesting_depth(html: &str, pos: usize) -> usize {
        let before = &html[..pos];
        let mut depth: usize = 0;
        let mut i = 0;
        while i < before.len() {
            let mut found = false;
            for pattern in &["class=\"comment", "class='comment", "id=\"comment-"] {
                if let Some(p) = find_ci(&before[i..], pattern) {
                    let p = i + p;
                    let tag_start = before[..p].rfind('<').unwrap_or(p);
                    if let Some(gt) = before[tag_start..].find('>') {
                        let tag = &before[tag_start..=tag_start + gt];
                        let tag_name = Self::extract_tag_name(tag);
                        let close_tag = format!("</{}>", tag_name);
                        let content_start = tag_start + gt + 1;
                        if let Some(end) = Self::find_matching_close(
                            before,
                            content_start,
                            &format!("<{}", tag_name),
                            &close_tag,
                        ) {
                            if end >= pos {
                                depth += 1;
                            }
                            i = end + close_tag.len();
                            found = true;
                            break;
                        }
                    }
                }
            }
            if !found {
                break;
            }
        }
        depth.saturating_sub(1)
    }

    /// Extract the value from a quoted attribute string like ="value" or ='value'.
    fn extract_quoted_value(s: &str) -> Option<String> {
        let mut i = 0;
        while i < s.len() && s.as_bytes()[i].is_ascii_whitespace() {
            i += 1;
        }
        let quote = *s.as_bytes().get(i)? as char;
        if quote != '"' && quote != '\'' {
            return None;
        }
        let val_start = i + 1;
        let val_end = s[val_start..].find(quote)? + val_start;
        Some(s[val_start..val_end].to_string())
    }

    /// Strip all HTML tags from a string, leaving only text.
    fn strip_tags(html: &str) -> String {
        let mut out = String::with_capacity(html.len());
        let mut in_tag = false;
        for c in html.chars() {
            if c == '<' {
                in_tag = true;
            } else if c == '>' {
                in_tag = false;
            } else if !in_tag {
                out.push(c);
            }
        }
        out
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn simple_paragraph() {
        let html = "<p>Hello world</p>";
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Hello world"));
    }

    #[test]
    fn heading_conversion() {
        let html = "<h1>Title</h1><h2>Subtitle</h2>";
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Title"));
        assert!(md.contains("Subtitle"));
    }

    #[test]
    fn removes_scripts_and_styles() {
        let html = r#"
            <html>
            <head><style>body{color:red}</style></head>
            <body>
                <script>alert('x')</script>
                <p>Content</p>
            </body>
            </html>
        "#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("alert"));
        assert!(!md.contains("color:red"));
        assert!(md.contains("Content"));
    }

    #[test]
    fn strips_images_when_false() {
        let html = r#"<p>Text before</p><img src="a.png" alt="pic"><p>Text after</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("a.png"));
        assert!(!md.contains("pic"));
        assert!(md.contains("Text before"));
        assert!(md.contains("Text after"));
    }

    #[test]
    fn keeps_images_when_true() {
        let html = r#"<p>Text before</p><img src="a.png" alt="pic"><p>Text after</p>"#;
        let md = PageToMarkdown::convert(html, true, false, false, &[]).unwrap();
        assert!(md.contains("a.png"));
        assert!(md.contains("pic"));
        assert!(md.contains("Text before"));
        assert!(md.contains("Text after"));
    }

    #[test]
    fn strips_self_closing_images() {
        let html = r#"<p>Before</p><img src="b.png" alt="self"/><p>After</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("b.png"));
        assert!(!md.contains("self"));
        assert!(md.contains("Before"));
        assert!(md.contains("After"));
    }

    #[test]
    fn strips_iframe_tags() {
        let html = r#"
            <p>Before</p>
            <iframe src="https://video.ibm.com/embed/123" allowfullscreen></iframe>
            <p>After</p>
        "#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("iframe"));
        assert!(!md.contains("video.ibm.com"));
        assert!(md.contains("Before"));
        assert!(md.contains("After"));
    }

    #[test]
    fn strips_iframe_tags_self_closing() {
        let html = r#"<p>Before</p><iframe src="map.html"/><p>After</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("iframe"));
        assert!(!md.contains("map.html"));
        assert!(md.contains("Before"));
        assert!(md.contains("After"));
    }

    #[test]
    fn strips_nav_tags() {
        let html = r#"<nav><a href="/">Home</a><a href="/about">About</a></nav><p>Content</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("Home"));
        assert!(!md.contains("About"));
        assert!(md.contains("Content"));
    }

    #[test]
    fn strips_footer_tags() {
        let html = r#"<p>Article</p><footer>Copyright 2025</footer>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Article"));
        assert!(!md.contains("Copyright"));
    }

    #[test]
    fn strips_aside_tags() {
        let html = r#"<p>Main text</p><aside>Sidebar content</aside>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Main text"));
        assert!(!md.contains("Sidebar"));
    }

    #[test]
    fn strips_noscript_tags() {
        let html = r#"<noscript>Please enable JS</noscript><p>Visible</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("enable JS"));
        assert!(md.contains("Visible"));
    }

    #[test]
    fn strips_form_tags() {
        let html = r#"<form action="/submit"><input type="text"/><button>Go</button></form><p>Text</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("submit"));
        assert!(!md.contains("button"));
        assert!(md.contains("Text"));
    }

    #[test]
    fn strips_html_comments() {
        let html = r#"<p>Before</p><!-- this is a comment --><p>After</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("comment"));
        assert!(md.contains("Before"));
        assert!(md.contains("After"));
    }

    #[test]
    fn strips_noise_tags_case_insensitive() {
        let html = r#"<NAV>Menu</NAV><p>Body</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("Menu"));
        assert!(md.contains("Body"));
    }

    #[test]
    fn preserves_content_between_noise_tags() {
        let html = r#"<nav>Nav</nav><p>First</p><footer>Foot</footer><p>Second</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("Nav"));
        assert!(!md.contains("Foot"));
        assert!(md.contains("First"));
        assert!(md.contains("Second"));
    }

    #[test]
    fn code_block_language_preserved() {
        let html = r#"<pre><code class="language-rust">fn main() {}</code></pre>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("```rust"), "expected language annotation, got: {}", md);
        assert!(md.contains("fn main()"));
    }

    #[test]
    fn code_block_language_python() {
        let html = r#"<pre><code class="language-python">print("hello")</code></pre>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("```python"), "expected language annotation, got: {}", md);
        assert!(md.contains("print"));
    }

    #[test]
    fn code_block_no_language_stays_plain() {
        let html = r#"<pre><code>plain code</code></pre>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("```"));
        assert!(!md.contains("```rust"));
        assert!(!md.contains("```python"));
        assert!(md.contains("plain code"));
    }

    #[test]
    fn multiple_code_blocks_with_languages() {
        let html = r#"<pre><code class="language-rust">let x = 1;</code></pre><p>text</p><pre><code class="language-go">fmt.Println()</code></pre>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("```rust"));
        assert!(md.contains("```go"));
        assert!(md.contains("let x = 1;"));
        assert!(md.contains("fmt.Println()"));
    }

    #[test]
    fn strips_header_tags_by_default() {
        let html = r#"<header><h1>Site Title</h1><nav>Menu</nav></header><p>Article body</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("Site Title"));
        assert!(!md.contains("Menu"));
        assert!(md.contains("Article body"));
    }

    #[test]
    fn keeps_header_when_requested() {
        let html = r#"<header><h1>Article Title</h1></header><p>Article body</p>"#;
        let md = PageToMarkdown::convert(html, false, true, false, &[]).unwrap();
        assert!(md.contains("Article Title"));
        assert!(md.contains("Article body"));
    }

    #[test]
    fn dedup_removes_duplicate_paragraphs() {
        let html = r#"<p>This is a long paragraph that should be deduplicated when it appears twice.</p><p>This is a long paragraph that should be deduplicated when it appears twice.</p><p>Unique content here.</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        let count = md.matches("deduplicated").count();
        assert_eq!(count, 1, "duplicate paragraph should appear once, got: {}", md);
        assert!(md.contains("Unique content"));
    }

    #[test]
    fn dedup_keeps_short_blocks() {
        let html = r#"<h1>Title</h1><h1>Title</h1><p>Body text</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Title"));
        assert!(md.contains("Body text"));
    }

    #[test]
    fn dedup_preserves_unique_blocks() {
        let html = r#"<p>First unique paragraph with enough text to exceed the threshold.</p><p>Second unique paragraph with different content entirely.</p><p>Third unique paragraph also different from the rest.</p>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("First unique"));
        assert!(md.contains("Second unique"));
        assert!(md.contains("Third unique"));
    }

    #[test]
    fn main_content_extracts_article_tag() {
        let html = r#"<nav>Home About</nav><article><h1>Article Title</h1><p>This is the main article content that should be extracted.</p></article><footer>Copyright 2025</footer>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("Article Title"));
        assert!(md.contains("main article content"));
        assert!(!md.contains("Copyright"));
    }

    #[test]
    fn main_content_extracts_main_tag() {
        let html = r#"<div>Sidebar noise</div><main><p>Main content goes here with enough text to be meaningful.</p></main><aside>Related links</aside>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("Main content"));
        assert!(!md.contains("Sidebar noise"));
    }

    #[test]
    fn main_content_extracts_role_main_div() {
        let html = r#"<div class="sidebar">Sidebar</div><div role="main"><p>This is the main content extracted via role attribute.</p></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("role attribute"));
        assert!(!md.contains("Sidebar"));
    }

    #[test]
    fn main_content_falls_back_to_full_html() {
        let html = r#"<div><p>Just some content without semantic tags.</p></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("Just some content"));
    }

    #[test]
    fn main_content_disabled_by_default() {
        let html = r#"<nav>Navigation</nav><article><p>Article content here.</p></article><footer>Footer</footer>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("Article content"));
    }

    #[test]
    fn readability_extracts_content_div_from_layout() {
        let html = r#"<div><a href="/">Home</a><a href="/about">About</a><a href="/contact">Contact</a><a href="/blog">Blog</a><a href="/shop">Shop</a></div><div><h2>Article Title</h2><p>This is a substantial article body with enough text content to score well in the readability algorithm. It contains meaningful paragraphs of text that should be extracted as the main content of the page, far exceeding the navigation links above.</p><p>Another paragraph with additional content to further increase the text density score of this content block compared to the navigation block which consists mostly of short link texts.</p></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("substantial article body"));
        assert!(!md.contains("Shop"));
    }

    #[test]
    fn readability_falls_back_when_no_semantic_tags() {
        let html = r#"<div><a href="/">Home</a><a href="/about">About</a></div><div><p>This is the main content paragraph with enough text to exceed the readability threshold for extraction. It has substantial body text that should be identified as the primary content block by the scoring algorithm.</p></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("main content paragraph"));
        assert!(!md.contains("About"));
    }

    #[test]
    fn readability_returns_full_html_when_no_good_candidate() {
        let html = r#"<div><p>Short.</p></div><div><p>Brief.</p></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("Short") || md.contains("Brief"));
    }

    #[test]
    fn readability_prefers_text_over_navigation() {
        let nav_html = r#"<div><a href="/a">Link A</a><a href="/b">Link B</a><a href="/c">Link C</a></div>"#;
        let content_html = r#"<div><p>This block has substantial text content that should score higher than the navigation block which only contains short link texts. The readability algorithm should correctly identify this as the main content area of the page.</p></div>"#;
        let html = format!("{}{}", nav_html, content_html);
        let md = PageToMarkdown::convert(&html, false, false, true, &[]).unwrap();
        assert!(md.contains("substantial text content"));
        assert!(!md.contains("Link A"));
        assert!(!md.contains("Link B"));
        assert!(!md.contains("Link C"));
    }

    #[test]
    fn readability_section_tag_supported() {
        let html = r#"<section><a href="/x">Short</a><a href="/y">Short2</a></section><section><p>This section contains the primary article content with enough text to be identified by the readability scoring algorithm as the main content block on the page, exceeding the navigation section in text density.</p></section>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("primary article content"));
        assert!(!md.contains("Short2"));
    }

    #[test]
    fn paragraph_readability_extracts_dense_cluster() {
        let html = r#"<div><a href="/">Home</a><a href="/about">About</a></div><p>Short nav text.</p><p>This is the first paragraph of the main article content with substantial text that should be captured by the paragraph-level readability scoring algorithm as part of a dense cluster.</p><p>Here is the second paragraph continuing the article with more meaningful content that contributes to the overall text density of this cluster of paragraphs.</p><p>The third paragraph adds even more substantial text content to ensure the sliding window picks up this cluster as the highest scoring region of the page.</p><p>Finally the fourth paragraph rounds out the content cluster with additional text that should push the combined score well above the threshold for extraction.</p><div><a href="/privacy">Privacy</a><a href="/terms">Terms</a></div>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("first paragraph"));
        assert!(md.contains("fourth paragraph"));
        assert!(!md.contains("Privacy"));
        assert!(!md.contains("Terms"));
    }

    #[test]
    fn paragraph_readability_falls_back_when_no_divs() {
        let html = r#"<p>Brief intro.</p><p>This is a substantial article paragraph with enough text content to be identified as the main content by the paragraph-level readability scoring algorithm when no div or section containers are present.</p><p>Another paragraph with additional content to build up the text density score of this cluster for extraction by the sliding window approach.</p><p>More content here to ensure the window score exceeds the threshold for meaningful extraction by the algorithm.</p>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("substantial article paragraph"));
    }

    #[test]
    fn paragraph_readability_skips_short_paragraphs() {
        let html = r#"<p>OK.</p><p>Sure.</p><p>Yep.</p><p>No.</p><p>Fine.</p>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        // All paragraphs are too short — should return full HTML
        assert!(md.contains("OK") || md.contains("Sure") || md.contains("Yep"));
    }

    #[test]
    fn paragraph_readability_extracts_best_window() {
        let html = r#"<p>Nav link one.</p><p>Nav link two.</p><p>Nav link three.</p><p>This paragraph contains the real article content that should be extracted by the paragraph readability algorithm because it has substantially more text than the navigation paragraphs above and below it.</p><p>Continuing the real article content with another substantial paragraph that should be part of the extracted window along with the previous paragraph.</p><p>Footer text one.</p><p>Footer text two.</p>"#;
        let md = PageToMarkdown::convert(html, false, false, true, &[]).unwrap();
        assert!(md.contains("real article content"));
    }

    #[test]
    fn comments_extracted_from_forum_page() {
        let html = r#"<html><head><title>Forum Thread</title></head><body>
            <h1>Discussion Topic</h1>
            <p>Original post content here.</p>
            <div class="comment" id="comment-1">
                <span class="author">Alice</span>
                <div class="comment-body"><p>First comment with some text.</p></div>
            </div>
            <div class="comment" id="comment-2">
                <span class="author">Bob</span>
                <div class="comment-body"><p>Second comment agreeing with Alice.</p></div>
            </div>
            <div class="comment" id="comment-3">
                <span class="author">Charlie</span>
                <div class="comment-body"><p>Third comment with a different perspective on the topic.</p></div>
            </div>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("## Comments"));
        assert!(md.contains("Alice"));
        assert!(md.contains("Bob"));
        assert!(md.contains("Charlie"));
        assert!(md.contains("First comment"));
        assert!(md.contains("Second comment"));
        assert!(md.contains("Third comment"));
    }

    #[test]
    fn comments_not_extracted_from_non_forum_page() {
        let html = r#"<html><head><title>Article</title></head><body>
            <h1>Regular Article</h1>
            <p>This is a normal article with no comments section.</p>
            <p>It has multiple paragraphs of content.</p>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("## Comments"));
    }

    #[test]
    fn comments_extracted_with_data_author() {
        let html = r#"<html><body>
            <h1>Thread</h1>
            <p>Post content.</p>
            <div class="comment" data-author="johndoe">
                <div class="comment-body"><p>Great post, thanks for sharing.</p></div>
            </div>
            <div class="comment" data-author="janedoe">
                <div class="comment-body"><p>I disagree with some points.</p></div>
            </div>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("## Comments"));
        assert!(md.contains("johndoe"));
        assert!(md.contains("janedoe"));
        assert!(md.contains("Great post"));
    }

    #[test]
    fn comments_not_extracted_with_only_one_comment() {
        let html = r#"<html><body>
            <h1>Page</h1>
            <p>Content.</p>
            <div class="comment" id="comment-1">
                <span class="author">Solo</span>
                <div class="comment-body"><p>Only one comment here.</p></div>
            </div>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(!md.contains("## Comments"));
    }

    #[test]
    fn comments_nested_with_indentation() {
        let html = r#"<html><body>
            <h1>Thread</h1>
            <p>Post.</p>
            <div class="comment" id="comment-1">
                <span class="author">Parent</span>
                <div class="comment-body"><p>Top level comment.</p></div>
                <div class="comment" id="comment-2">
                    <span class="author">Child</span>
                    <div class="comment-body"><p>Reply to parent.</p></div>
                </div>
            </div>
            <div class="comment" id="comment-3">
                <span class="author">Another</span>
                <div class="comment-body"><p>Another top level comment.</p></div>
            </div>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("## Comments"));
        assert!(md.contains("Parent"));
        assert!(md.contains("Child"));
        assert!(md.contains("Top level comment"));
        assert!(md.contains("Reply to parent"));
    }

    #[test]
    fn comments_extracted_from_reddit_style() {
        let html = r#"<html><body>
            <h1>Reddit Post</h1>
            <p>Post content.</p>
            <div class="comment" data-testid="comment-1">
                <div data-author="redditor1">
                    <div class="md"><p>Reddit style comment.</p></div>
                </div>
            </div>
            <div class="comment" data-testid="comment-2">
                <div data-author="redditor2">
                    <div class="md"><p>Another Reddit comment.</p></div>
                </div>
            </div>
        </body></html>"#;
        let md = PageToMarkdown::convert(html, false, false, false, &[]).unwrap();
        assert!(md.contains("## Comments"));
        assert!(md.contains("redditor1"));
        assert!(md.contains("Reddit style comment"));
    }

    #[test]
    fn absolutize_links_converts_relative_to_absolute() {
        let md = "[About](/about) [Contact](/contact-us)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/page");
        assert!(result.contains("https://example.com/about"));
        assert!(result.contains("https://example.com/contact-us"));
    }

    #[test]
    fn absolutize_links_handles_protocol_relative() {
        let md = "[Link](//cdn.example.com/file)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/page");
        assert!(result.contains("https://cdn.example.com/file"));
    }

    #[test]
    fn absolutize_links_leaves_absolute_unchanged() {
        let md = "[Link](https://other.com/page)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/page");
        assert!(result.contains("https://other.com/page"));
    }

    #[test]
    fn absolutize_links_leaves_anchor_links_unchanged() {
        let md = "[Section](#section)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/page");
        assert!(result.contains("#section"));
    }

    #[test]
    fn absolutize_links_resolves_relative_paths() {
        let md = "[Prev](../parent) [Next](./child)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/blog/post");
        // ../parent from /blog/post → /parent (go up from post to blog, then up from blog to root)
        assert!(result.contains("https://example.com/parent"), "got: {}", result);
        // ./child from /blog/post → /blog/child (same directory as post)
        assert!(result.contains("https://example.com/blog/child"), "got: {}", result);
    }

    #[test]
    fn absolutize_links_handles_image_links() {
        let md = "![Photo](/images/photo.jpg)";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com/page");
        assert!(result.contains("https://example.com/images/photo.jpg"));
    }

    #[test]
    fn absolutize_links_preserves_surrounding_text() {
        let md = "Hello [world](/world) and [universe](/universe) end";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com");
        assert!(result.starts_with("Hello "));
        assert!(result.contains("https://example.com/world"));
        assert!(result.contains("https://example.com/universe"));
        assert!(result.ends_with("end"));
    }

    #[test]
    fn absolutize_links_invalid_base_returns_original() {
        let md = "[Link](/path)";
        let result = PageToMarkdown::absolutize_links(md, "not-a-url");
        assert_eq!(result, md);
    }

    #[test]
    fn absolutize_links_preserves_unicode_text() {
        let md = "[リンク](/page) 日本語テキスト";
        let result = PageToMarkdown::absolutize_links(md, "https://example.com");
        assert!(result.contains("https://example.com/page"));
        assert!(result.contains("日本語テキスト"));
    }

    #[test]
    fn exclude_selector_class_removes_element() {
        let html = r#"<p>Keep this</p><div class="ad">Advertisement</div><p>Also keep</p>"#;
        let selectors = vec![".ad".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep this"));
        assert!(md.contains("Also keep"));
        assert!(!md.contains("Advertisement"));
    }

    #[test]
    fn exclude_selector_id_removes_element() {
        let html = r#"<p>Keep this</p><div id="sidebar">Sidebar content</div><p>Also keep</p>"#;
        let selectors = vec!["#sidebar".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep this"));
        assert!(md.contains("Also keep"));
        assert!(!md.contains("Sidebar"));
    }

    #[test]
    fn exclude_selector_multiple_selectors() {
        let html = r#"<div class="ad">Ad</div><p>Keep</p><div id="promo">Promo</div><p>Also keep</p>"#;
        let selectors = vec![".ad".to_string(), "#promo".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep"));
        assert!(md.contains("Also keep"));
        assert!(!md.contains("Ad"));
        assert!(!md.contains("Promo"));
    }

    #[test]
    fn exclude_selector_class_word_boundary() {
        let html = r#"<div class="ad-banner">Banner</div><p>Keep</p><div class="notad">Should keep</div>"#;
        let selectors = vec![".ad".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep"));
        assert!(md.contains("Banner"));
        assert!(md.contains("Should keep"));
    }

    #[test]
    fn exclude_selector_class_multi_class_match() {
        let html = r#"<div class="box highlight">Highlighted</div><p>Keep</p>"#;
        let selectors = vec![".highlight".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep"));
        assert!(!md.contains("Highlighted"));
    }

    #[test]
    fn exclude_selector_empty_does_nothing() {
        let html = r#"<p>Keep this</p><div class="ad">Ad</div>"#;
        let selectors: Vec<String> = vec![];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep this"));
        assert!(md.contains("Ad"));
    }

    #[test]
    fn exclude_selector_nested_elements_removed() {
        let html = r#"<div class="ad"><h2>Ad Title</h2><p>Ad content</p></div><p>Keep this</p>"#;
        let selectors = vec![".ad".to_string()];
        let md = PageToMarkdown::convert(html, false, false, false, &selectors).unwrap();
        assert!(md.contains("Keep this"));
        assert!(!md.contains("Ad Title"));
        assert!(!md.contains("Ad content"));
    }
}